This is an application for the Tumor Microenvironment Network from a highly collaborative group of investigators from Columbia University who have worthed together for over 5 years on the tumor microenvironment. The focus of the program is the role of myofibroblasts (MF)/cancer-associated fibroblasts (CAF) in digestive cancers, and includes studies of gastric, hepatocellular and pancreatic cancer. The three cancers are among the most common and lethal of solid tumors, and evidence suggests that each is strongly promoted at early stages by alpha-SMA+ fibroblastic cells. In addition, recent work suggests that Helicobacter-dependent gastric cancer is promoted by CAFs that are derived from bone marrow derived mesenchymal stem cell (MSCs) and that contribute to the MSC and tumor niche; that hepatocellular cancer is promoted by myofibroblasts that are activated by the microbiome and TLR4 signaling; and that CAFs in digestive cancers show early global DNA hypomethylation with focal gains or promoter methylation, that results in a unique susceptibility to further loss of DNA methylation. Based on these and other observations, three highly interactive projects are proposed. Project 1 (Wang) will further explore the MSC origin of gastric cancer MF/CAFs, their interaction with microbiome signals, their contribution to cancer growth, and the mechanism of hypomethylation; Project 2 (Schwabe) will explore the role of damage-associated molecular patterns (DAMPs), TLR4, and the microbiome to the tumor-promoting myofibroblast niche; Project 3 (Tycko) will examine the relationship of pancreatic and hepatocellular CAF hypomethylation to changes in gene expression and in vivo behavior, and examine the effects of decitabine or conditional Dnmtl KO in CAFs on prevention of tumor progression. The three projects will jointly address the three major themes which are the origins of CAFs, their activation by the gut microbiome, and the role of DNA hypomethylation. The three projects are supported by an Administrative/Bioinformatics Core and by a Germ-Free Mouse Core at MIT. The three projects are highly translational with heavy use of human tissues and physiological mouse models of digestive cancers, and are linked to a Phase 1 clinical trial in human patients with pancreatic cancer.